Modeling Of Multi-objective Human-robot Parallel Disassembly Line Balancing Problem And Hybrid Teaching Learning Based Optimization Algorithm

Facing the increasing number of waste products,The efficient and feasible disassembly methods are of great significance for resource recovery and remanufacturing.Although the worker disassembly has the advantages of high flexibility and quick response to emergencies.However,the worker disassembly has low efficiency,high labor cost,and easy fatigue,which cannot meet people’s increasing demand for disassembly efficiency,and cannot avoid the harm to human caused by disassembling harmful parts.In view of the shortcomings of worker disassembly,the robot disassembly can achieve standardized operations,high production efficiency,save labor,and not easy to fatigue.However,the robot disassembly ability is limited due to technical limitations.It cannot overcome the uncertainty of the structure and failure form of the product,the flexibility is poor.According to the above actual situation,combining the advantages and disadvantages of worker disassembly and robot disassembly,it is proposed to study human-robot disassembly line balancing problem on the basis of parallel layout.The disassembly work is labor-intensive.In order to improve disassembly efficiency,the single-operator human-robot disassembly line which uses robots to replace some disassembly workers is addressed.Besides,aiming at the hazardous parts and complex parts with different properties contained in the waste products,the multi-operator human-robot disassembly line is also addressed in which the robots are arranged to disassemble hazardous parts and the workers are arranged to disassemble complex parts.Considering that the parallel disassembly line has the advantages of being able to disassemble multiple products at the same time with high flexibility and high disassembly efficiency,the human-robot disassembly line with parallel line layout is proposed.And the mixed integer programming models for the single-operator humanrobot parallel disassembly line balancing problem and the multi-operator human-robot parallel disassembly line balancing problem are constructed respectively,which make up for the deficiency of the conceptual models for the parallel disassembly line balancing problem and human-robot disassembly to a certain extent.The multi-operator human-robot parallel disassembly line balancing problem is also an extended problem of the single-operator humanrobot parallel disassembly line balancing problem,in which operator allocation constraints,time relationship constraints and specific attribute task constraints are considered according to the characteristics of the problem.The LINGO solver is used to solve the small-scale examples of the two problems accurately to verify the correctness of the models.In order to effectively solve the two kinds of human-robot parallel disassembly line balancing problem,a hybrid teaching optimization algorithm is presented.The cross operator and the exchange operator are applied to simulate the teaching operation and mutual learning operation of the teaching optimization algorithm.And the self-study operation based on mutation operator is introduced into the original teaching optimization algorithm to increase the diversity of solutions.The Metropolis criterion of simulated annealing algorithm is improved to calculate the screening probability which is used to accept the the inferior solution to improve the local search ability of the algorithm.Two novel decoding method are designed for the two kinds of problems respectively according to the characteristics of the problems.The performance of the proposed algorithm is verified by testing the P25 disassembly example,P52 disassembly example and 19 benchmark disassembly examples,in which orthogonal experiments are carried out to determine the optimal parameters of the algorithm for each disassembly example.Finally,the waste LCD TV and CRT TV these two kinds of TV sets are taken as the application instance,and the proposed hybrid teaching optimization algorithm and other three comparison algorithms including including artificial fish swarm algorithm,genetic algorithm and particle swarm algorithm are used to solve the single-operator and the multi-operator human-robot parallel disassembly line balancing problem for the two kinds of TV sets.The solution results of the proposed hybrid teaching optimization algorithm are compared with the other three algorithms respectively,and the comparison resultsidentify the superiority of the proposed algorithm,through which multiple non-inferior solution schemes are available for decision makers to choose.Besides,it can be concluded that the multi-operator human-robot parallel disassembly line can effectively reduce the number of workstations opened and the disassembly cost,while the workload of the workstations of single-operator human-robot parallel disassembly line is more balanced.